Study On Dysarthria Of Parkinson’s Disease Based On Co-occurrence Direction Attribute Topology

Parkinson’s disease is a common neurodegenerative disease.As one of the early typical symptoms,dysarthria has been widely concerned.In recent years,with the development of formal structure representation methods represented by attribute topology,disease analysis methods based on formal structure have become one of the hottest spots in the field of disease research.Aiming at the application of Parkinson’s disease speech signal in formal structure,this paper starts a series of researches on Parkinson’s disease speech from the perspective of attribute topology.Firstly,this paper proposes a method to turn speech signal into attribute topology.In the process of conversion,the speech is transformed by short-time Fourier transform to get the time-frequency form.In order to describe the relationship between direction and energy point in spectrogram,a direction statistics method based on different time-domain mixing ratio is proposed to obtain more complete information;For the relationship between direction and energy point,a formal context establishment method based on kernel density estimation is proposed;According to the formal context,the co-occurrence attribute topology description method is proposed,which realizes the establishment of the cooccurrence direction attribute topology of the Parkinson’s disease speech.Secondly,by analyzing the whole structural characteristics of attribute topology,a structural feature extraction method is proposed.According to the structural characteristics of attribute topology,a connected component calculation algorithm is proposed to realize the statistics of the number of connected components,thereby obtaining structural statistical features,and the dimension reduction and classification of the features are carried out;Experiments show that the structural feature extraction method can complete the Parkinson’s disease speech classification.Thirdly,by analyzing the time-frequency characteristics of attribute topology,a timefrequency feature extraction method is proposed.Aiming at the time domain and frequency domain characteristics of Parkinson’s disease speech,the spectrogram is processed by sliding window,and the time domain features of the time domain information and the frequency domain features of the frequency domain information are obtained respectively,and the dimension reduction and classification of the features are carried out;Experiments show that the frequency information in speech is an important factor affecting the classification.Finally,by analyzing the interaction characteristics between nodes,a co-occurrence feature extraction method is proposed.According to the description of co-occurrence strength relationship between nodes in attribute topology structure,the whole co-occurrence statistical features of structure are obtained,and the dimension reduction and classification of the features are carried out;Experiments show that co-occurrence features can well describe the pathological characteristics of Parkinson’s disease speech.